JPH09245271A - Object detection system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect the size of a travelling object by obtaining difference between prescribed fields or frames with respect to images which are picked-up be means of two TV cameras arranged at a prescribed interval and obtaining the size of the object based on difference data and visual field angle, etc. SOLUTION: The two TV cameras 1 and 3 are arranged at distance d2 . The optical axis 0 of the camera 1 is in parallel to the optical axis 0' of the camera 3. The image picked-up by the camera 1 is written in an image memory 2 and the image picked-up by the cameras 3 is written in the image memory 4. A controller 5 obtains difference between the prescribed fields or the frames concerning the images picked-up by the respective cameras and a processing for obtaining the size of the object is executed based on the difference data and the visual field angle of the respective cameras 1 and 3, etc. Thus, the size of the travelling object such as an intruder is detected.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object detection system for detecting the size of a moving object.

[0002]

2. Description of the Related Art
In most cases, a heat ray sensor is used for a crime prevention system for detecting an intruder and a so-called door switch for controlling opening / closing of an automatic door.

However, since the heat ray sensor detects heat rays, there is a problem that not only humans but also small animals such as rats are detected.

The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide an object detection system capable of detecting the size of a moving object.

[0005]

In order to achieve the above object, the object detection system of the present invention comprises two television cameras arranged at a predetermined interval and images taken by the respective television cameras. About the difference between predetermined fields or frames, and their difference data,
And a control means for performing a process for obtaining the size of the object based on the viewing angle of each television camera.

[0006]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an embodiment of an object detection system according to the present invention, in which 1 is a television camera (hereinafter simply referred to as a camera), 2 is an image memory, 3 is a camera, 4 is an image memory, Reference numeral 5 indicates a control device.

The two cameras 1 and 3 are arranged with a distance d ₂ therebetween. It is assumed here that the optical axis O of the camera 1 and the optical axis O ′ of the camera 3 are parallel to each other for easy understanding.

The image captured by the camera 1 is stored in the image memory 2
Is written to. The image memory 2 has a capacity capable of writing images of a plurality of frames, and writing is controlled by the control device 5. Similarly, the image captured by the camera 3 is written in the image memory 4. The image memory 4 has a capacity capable of writing a plurality of frames of image data, at least three frames of image data, and the writing is controlled by the control device 5.

When a moving object exists in the images picked up by the cameras 1 and 3, the control device 5 performs a process of detecting the size of the moving object. The process is as follows.

Now, as shown in FIG. 2A, it is assumed that there is an elliptical object at a position indicated by S in a certain frame imaged by the camera 1 and the object has moved to a position indicated by S'in the next frame. Then, when the image difference between these frames is calculated, the shaded portion in FIG. 2A is obtained as a significant difference. Here, the significant difference means that the value of the inter-frame difference is equal to or larger than a predetermined threshold value, that is, the low-level noise is excluded.

Next, as shown in FIG. 2B, the control device 5 obtains a circumscribed rectangle 6 of the obtained region of significant difference, and sets this circumscribed rectangle 6 as a moving object. Also, this circumscribed rectangle 6
The center of is the center of gravity D of the moving object. Although the above is described with respect to the camera 1, the same applies to the camera 3.

Then, the control unit 5 obtains the horizontal size and the vertical size of the circumscribed rectangle 6.

Of course, since the circumscribing rectangle 6 does not represent an actual moving object itself, the horizontal size and the vertical size of the circumscribing rectangle 6 are different from the size of the actual moving object. What is important in the system is to recognize humans and small animals. Moreover, since the moving speed of humans can be 3 m / sec or less at the maximum, the size of the circumscribing rectangle 6 in the horizontal direction and the vertical direction can be determined. It is possible to judge whether the detected moving object is a human or a small animal.

As described above, the controller 5 stores the image data of a certain frame from the camera 1 in the image memory 2.
, The image data of the next frame is written to the second area of the image memory 2, and the significant difference image data of the difference between the frames is written to the third area of the image memory 2.
In the area, the circumscribed rectangle 6 is written in the significant difference image data, the significant difference image data is erased, and the center of gravity D is written in the center of the circumscribed rectangle 6. As a result, image data in which only the circumscribing rectangle 6 and its center of gravity D are written is obtained in the third area of the image memory 2 (hereinafter, this image data will be simply referred to as difference image data).

The above is the description of the camera 1 system.
The control device 5 performs the same process for the system of the camera 3.

In the above description, the difference between frames is taken, but this is for easier understanding, and the difference between predetermined fields may be taken or the difference between a plurality of frames may be taken. It will be apparent to those skilled in the art.

When the differential image data is written in the image memories 2 and 4 as described above, the control device 5 determines the horizontal size and the vertical direction of the circumscribed rectangle 6 based on these two differential image data. The process of finding the size of is performed.

The principle is as follows. As shown in FIG. 1, two cameras 1 and 3 are used in this object detection system. Here, the detection of the horizontal size of the circumscribing rectangle 6 for the camera 1 will be examined as follows. .

FIG. 3 is a ray diagram showing the relationship between the surface on which the object is present (hereinafter referred to as the object surface) P and the image pickup surface Q of the camera 1, where L is the center plane of the lens of the camera 1. (Hereinafter, simply referred to as the center plane), O indicates the optical axis. Therefore, the image memory 2 stores the data of the image projected on the imaging surface Q.

When an image of a certain subject is taken by the camera 1, a subject plane parallel to the image pickup plane of the camera 1 can be taken at the position of the subject. Therefore, as shown in FIG. 3, it is assumed that there is a subject H having a width a ₁ on the subject surface P and the width of the subject image I on the imaging surface Q is a ₂ . In the figure, b ₁ is the distance between the center plane L and the object plane P, b ₂ is the distance between the center plane L and the imaging plane Q, c ₂ is the width of the imaging plane Q, and c ₁ is the imaging plane Q The width of the subject plane P to be displayed is shown. Further, e ₁ indicates the distance from the optical axis O to the center of the subject H. Further, β ₁ is the viewing angle of the lens, and γ ₁ is the angle between the optical axis O and the line connecting the center of the subject H and the center of the center plane L.

Since the shape of the subject H on the subject plane P and the subject image I on the imaging plane Q are similar, a ₁ : c ₁ = a ₂ : c ₂ (1) holds, and since FIG. 3, _{c 1 = 2 × b 1 ×} tan (β 1/2) is a ... (2), (1) , a 1 = (a 2 / c 2) × 2 × b 1 from (2) _{× tan (β 1/2)} ... (3) but is obtained, wherein, beta ₁ is known since it is the viewing angle of the lens camera 1 and a ₂ / c ₂ image memory 2
Can be obtained from the difference image data of That is, here, the size a _{1 of the} subject H is the width from end to end of the object when the object horizontally moves between frames, and the size a ₂ of the subject image I on the imaging surface Q corresponding to this is a ₂ Is the horizontal size of the circumscribing rectangle 6 in the differential image data, the value of a ₂ / c ₂ is the circumscribing rectangle with respect to the horizontal size c ₂ of the area in the image memory 2 in which the differential image data is written. It can be obtained as a ratio of the horizontal size a ₂ of 6.

Therefore, the size a _{1 of the} subject H can be obtained if the distance b ₁ between the center plane L and the subject plane P is known. The above is the description of the detection of the horizontal size of the circumscribing rectangle 6 for the camera 1.
The same applies to the detection of the vertical size of the circumscribing rectangle 6 for the camera 1.

Then, the method of obtaining the distance b ₁ between the center plane L and the object plane P will be described below.

By the way, the positional relationship in the horizontal direction regarding the camera 1, the camera 3, and the center of gravity D of the subject is shown in FIG. In FIG. 4, point A is the position of the camera 1, point B is the position of the camera 3, Q is the plane connecting the positions A of the camera 1 and the position B of the camera 3, P is parallel to the straight line Q through the center of gravity D of the subject. Shows the other side. In the figure, O is the optical axis of the camera 1, O'is the optical axis of the camera 3, β ₁ is the viewing angle of the camera 1,
β ₂ is the viewing angle of the camera 3, θ ₁ is the angle between the straight line connecting the position A of the camera 1 and the center of gravity D of the subject and the surface Q, θ ₂
Is a straight line connecting the position B of the camera 3 and the center of gravity D of the subject,
The angle formed by the surface Q, d ₂ , represents the distance between the camera 1 and the camera 3.

Now, in FIG. 4, the camera 1 to be obtained
The distance b ₁ between the object and the center of gravity of the object is the distance between the plane Q and the plane P. In FIG. 4, assuming that the position A of the camera 1 is the origin of the XY coordinates and the optical axis O of the camera 1 is the Y axis. , B ₁ = tan θ ₁ · X (4) holds for the camera 1 and the center of gravity D of the subject, and similarly b ₁ = tan θ ₂ · (X−d ₂ ) (5) for the center of gravity D of the camera 3 and the subject. Holds.

Since X = b ₁ / tan θ ₁ (6) according to the equation (4), when substituting this into the equation (5) and rearranging, b ₁ = tan θ ₁ · tan θ ₂ · d ₂ / (Tan θ ₂ −tan θ ₁ ) (7)

Here, d ₂ is the distance between the camera 1 and the camera 3 and is known. Therefore, if θ ₁ and θ ₂ are known, b ₁ is obtained and the value is substituted into the equation (3). By doing so, the size a ₁ of the subject in the horizontal direction is obtained.

There are the following two methods for obtaining these θ ₁ and θ ₂ . First, the first method is as follows. In this case, the cameras 1 and 3 are fixed, and the viewing angles of the cameras 1 and 3 are constant. Therefore, depending on the position of the center of gravity D of the subject in the difference image data, θ
₁ and θ ₂ are uniquely determined. This is clear.

Therefore, with respect to the camera 1, the control device 5 is provided with a table (hereinafter referred to as a first table) in which the value of the angle θ ₁ with respect to the position in the difference image data of the center of gravity D of the object is written, and the camera 3 is also provided. Regarding this, the control device 5 is provided with a table (hereinafter, referred to as a second table) in which the value of the angle θ ₂ with respect to the position of the center of gravity D of the subject in the difference image data is written.

In this case, the control device 5 obtains the angle θ ₁ by referring to the first table according to the position of the center of gravity D of the subject obtained from the difference image data of the image memory 2, and similarly the difference image data of the image memory 4 is obtained. The angle θ ₂ is obtained by referring to the second table according to the position of the center of gravity D of the subject obtained from the above, and these values of θ ₁ and θ ₂ are substituted into the equation (7) to obtain b ₁ and then b The value of ₁ is substituted into the equation (3) to perform the calculation for obtaining the horizontal size a ₁ of the circumscribed rectangle 6.

[0031] The value of the angle theta ₁ with respect to the position in the difference image data of the center of gravity D of the object to be written to the first table is previously determined angle theta ₁ with respect to every single pixel position of the difference image data Although it is desirable to write the difference image data in this case, since the amount of data to be written to the first table is large in this case, the difference image data is divided into appropriate regions in a matrix shape, and The value of the angle θ ₁ may be written. This also applies to the second table.

In addition, in the object detection system as shown in FIG. 1, the camera 1 and the camera 3 are generally adjustable in angle. Therefore, as the first table, the adjustable angle range is divided at a predetermined angle, and the first table is provided for each of the divided angle ranges, depending on the actually set angle. First of
It suffices to set whether to use the table in the control device 5. This also applies to the second table.

The above is the first method, but the second method is as follows. In FIG. 3, the above equation (2) holds, and in FIG. 4, the following equation (8) holds.

E ₁ = b ₁ · tan γ (8) Here, γ = 90−θ ₁ .

3 and 4 are views showing the same state from different viewpoints, the equations (2) and (8) hold at the same time.

[0036] Therefore, (2) of both sides respectively divided by the equation (8) of both sides and rearranging _{tanγ = 2 · tan (β 1} /2) · e 1 / c 1 ... (9) , and the thus γ since ^{_{= 90-θ 1 = tan -1}} {2 · tan (β 1/2) · e 1 / c 1} ... a (10), theta ₁ will be can be calculated by (11) below .

[0037] ^{_{θ 1 = 90-tan -1 {}} 2 · tan (β 1/2) · e 1 / c 1} ... (11) In this equation (11), beta ₁ is known, also, e ₁ /
The value of c ₁ can be obtained from the difference image data. This is because the shape of the subject and the subject image on the imaging surface of the camera 1 are similar to each other as described above. Therefore, when the difference image data is as shown in FIG. The horizontal distance e ₂ from the center of gravity D of the subject in the difference image data corresponds to e ₁ in FIG.
Therefore, the ratio between the horizontal sizes c ₂ and e ₂ of the difference image data becomes equal to the ratio between c ₁ and e ₁ in FIG.

Although the calculation of θ ₁ for the system of the camera 1 has been described above, the same applies to the calculation of θ ₂ for the system of the camera 3, and the system of the camera 3 corresponds to c ₁ and e ₁ in FIG. c ₁ values respectively ', e _1' When, can be obtained by ^{_{θ 2 = 90-tan -1 {}} 2 · tan (β 2/2) · e 1 '/ c 1'} ... (12).

Therefore, in this case, the control device 5 obtains θ ₁ based on the differential image data of the image memory 2 and the viewing angle β ₁ of the camera 1, and similarly, the differential image data of the image memory 4 and the camera 3 are obtained. Determine θ ₂ based on the viewing angle β ₂ ,
Then, by substituting these values of θ ₁ and θ ₂ into equation (7), b ₁
Then, the value of b ₁ is substituted into the equation (3) to perform a calculation for obtaining the horizontal size a ₁ of the circumscribed rectangle 6.

Although the case where the horizontal size of the circumscribing rectangle 6 is obtained has been described above, it is clear that the vertical size of the circumscribing rectangle 6 can be similarly obtained.

As described above, the control device 5 first obtains the inter-frame difference for each system of the camera 1 and the camera 3, obtains the difference image data based on the difference, and then the first method or By the second method, θ ₁ ,
seeking theta _2, and these theta _1, obtains a b ₁ by substituting the value of theta ₂ in (7), horizontal circumscribed rectangle 6 further assigns the value of the b ₁ (3) to the equation of the determined size a _1, then similarly determine the vertical size of the circumscribed rectangle 6.

If the obtained horizontal size and vertical size are both within preset values, it is determined that the moving object is a human, and a human is detected. Output a signal.

The processing for detecting the size of the circumscribed rectangle 6 described above is preferably performed for each frame. However, when the above calculation processing requires time, the size of the circumscribed rectangle 6 for one difference image data is determined. The process for obtaining the difference image data may be performed after the process is performed, and the size of the circumscribed rectangle 6 for the difference image data may be obtained.

It should be noted that there may be a plurality of moving objects within the field of view of the cameras 1 and 3, but in this case, since a plurality of circumscribing rectangles can be obtained in one difference image data, each circumscribing rectangle is On the other hand, the size in the horizontal direction and the size in the vertical direction are obtained, and if even one of them can be determined to be a human, a signal indicating that a human has been detected may be output.

It is obvious that this object detection system can be used as a security system for detecting an intruder or a door switch for controlling opening / closing of an automatic door.

Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment, and it is needless to say that various modifications can be made.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of an object detection system according to the present invention.

FIG. 2 is a diagram for explaining a circumscribing rectangle 6 and its center of gravity D obtained from inter-frame differences.

FIG. 3 is a ray diagram showing a relationship between a subject plane P and an imaging plane Q of the camera 1.

FIG. 4 is a diagram showing a positional relationship in the horizontal direction regarding a camera 1, a camera 3, and a center of gravity D of a subject.

FIG. 5 is a diagram for explaining an example of difference image data.

[Explanation of symbols]

1 ... Television camera, 2 ... Image memory, 3 ... Television camera, 4 ... Image memory, 5 ... Control device.

Claims (1)

[Claims] 1. Two TV cameras arranged at a predetermined interval, and a difference between predetermined fields or frames of images picked up by the respective TV cameras is calculated, and the difference data thereof, each TV. An object detection system, comprising: a control unit that performs a process of obtaining the size of an object based on a viewing angle of a camera.